Fix min_ident_chars: ignore on trait impl.

[t-webber]

fixes #13396

changelog: [min_ident_chars]: ignore lint when implementing a trait, to respect [renamed_function_params]

[rustbot]

r? @dswij

rustbot has assigned @dswij.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[github-actions]

Lintcheck changes for e8db4aa

Lint	Added	Removed	Changed
clippy::min_ident_chars	89	2454	39

This comment will be updated if you push new changes

[samueltardieu]

For a local trait, it should warn about the short parameter name in the trait method definition. I don't see a test for this, and I suspect it won't warn. To suppress warnings about trait implementations, we should ensure it warns at the trait definition so that users are warned at least once about the usage of short identifiers.

[t-webber]

Yes, I didn't look at that case, and it seems it never fired the lint even before my PR:

#![allow(dead_code)]
#![warn(clippy::min_ident_chars)]

trait Trait {
    fn func(x: u32) -> u32;
}

struct Struct;

impl Trait for Struct {
    fn func(x: u32) -> u32 {
        x
    }
}

fn main() {}

This code doesn't produce any warnings.

For functions with one char, it works though, and it warns on trait definition only, so I guess I should do the same thing for one-char-long params

[t-webber]

My previous message wasn't correct, as x is allowed as a single char by default. If you change my code and put another variable than x, the lint fires. It lints only on the implementation of the function, not the definition though.

Strange thing that the CIs stopped passing on an error in the source file, that I didn't change on this commit

[samueltardieu]

Could you add a "Limitations" section to the lint definition, after the "Example"? Something like

    /// ### Limitations
    /// Trait implementations which use the same function or parameter name as the trait declaration will
    /// not be warned about, even if the name is below the configured limit.

Also, it would be great to open an issue with label C-enhancement suggesting to add a new configuration option for this lint so that it triggers when using short parameter names even in the case of trait implementation: while we cannot do anything about the function names themselves, we can suggest renaming the function parameters in the implementation.

[samueltardieu]

r? samueltardieu
@rustbot author

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[t-webber]

Also, it would be great to open an issue with label C-enhancement suggesting to add a new configuration option for this lint so that it triggers when using short parameter names even in the case of trait implementation: while we cannot do anything about the function names themselves, we can suggest renaming the function parameters in the implementation.

cf. #15365

[t-webber]

@rustbot ready

[samueltardieu]

It is starting to look good. Could you test the suggested change (and added tests), and tell me what you think about it? Also, you may squash the commits into one, that will be necessary before merging.

[t-webber]

@rustbot ready

[samueltardieu]

Thanks. That makes it a large change, with many more changes in output too, but for a restriction lint I think this is fine.

[t-webber]

Thank you for your help!